Step 2 — Test set

Pick or upload the labelled dataset Luna Studio will evaluate the fine-tuned metric against.

The test set is the ground truth for the run. After training, Luna Studio scores the resulting metric against this dataset and reports F1, AUC-ROC, and other result metrics on the Run details page.

Step 2 — pick an existing test set or add a new one

Pick an existing test set

The Test set select shows test sets you’ve already added to this org. Each option includes a row count and source label, e.g. rag-eval-dataset-v2 — 320 rows · Uploaded.

Type into the select to filter by name.

Add a new test set

If you don’t have one yet, click the dropdown’s Add new test set action. The Add test set modal opens.

Add test set modal opened from Step 2 — same flow as the Datasets page

The modal title is Add test set with the subtitle “Test sets are curated labelled examples used to evaluate your metric.” Three sources are available — see Add a dataset for a complete reference.

Upload from local

Drag-and-drop a .csv or .jsonl file.

Fetch from URL

Paste an http://, https://, s3://, or gs:// URL.

Import from Galileo

Browse datasets in your connected Galileo workspace.

Required columns: depends on the metric’s input type, for more details see Prerequisites.

Importing from Galileo requires an active Galileo integration. If one isn’t configured, Luna Studio prompts you to add it inline before the import panel appears.

Validation

Luna Studio runs validation on the test set to ensure it meets the required schema / format / content rules. If there are any validation errors, they will be highlighted (See example below).

Step 2 with a validation error — the flow blocks Next until the schema is valid

Dataset preview

If validation completes, you should see a preview of the test set rows.

The preview is paginated so you can inspect rows without leaving the run creation flow.

Use the Calculate F1 score button to see the F1 score of your LLM-as-judge prompt on the selected test set. This is your benchmark score which Luna Studio will aim to achieve with a Luna metric.

Where to go next

Step 3 — Training set

Generate from the test set, or upload your own.

Add a dataset

Reference for all three dataset sources.